Best AI papers explained

We discuss the concept of textual gradient-based optimization for Large Language Model (LLM) workflows , contrasting it with traditional embedding space methods like soft prompt tuning. They introduce TextGrad as a foundational framework and focus heavily on LLM-AutoDiff as an advanced system designed to optimize complex, multi-component LLM applications represented as graphs, including features l...

This academic paper explores the theoretical underpinnings of large language models (LLMs) , particularly their generalization abilities. The authors propose an equivalence between autoregressive transformer-based LLMs and finite-state Markov chains as a framework for analysis. They use this framework to examine LLM inference, generalization during pre-training on dependent data, and in-context le...

This research explores Chain-of-Thought (CoT) reasoning in large language models by viewing it as a metastable Markov process . The authors model easy reasoning steps as dense clusters and hard steps as sparse connections, proving that search strategies rewarding these sparse edges improve efficiency by reducing the time to navigate between concept clusters. The study demonstrates that information...

This research explores how transformers adapt to changing causal structures in data , which is crucial for understanding their success in language processing. They introduce a new test using interleaved Markov chains with varying "lags" or dependencies. The paper shows that a three-layer transformer can learn to identify the correct lag and predict the next token , a process termed selec...

This document studies how transformers learn to predict sequential patterns in context , focusing on Markov Chains , a fundamental type of sequence. The research introduces a task called ICL-MC to investigate this, where models must learn from input sequences generated by different Markov Chains. The findings indicate that transformers develop "statistical induction heads" capable of cal...

This research investigates how transformers learn causal structure through gradient descent , focusing on their ability to perform in-context learning. The authors introduce a novel task involving random sequences with latent causal relationships and analyze a simplified two-layer transformer architecture. They demonstrate theoretically that gradient descent on the first attention layer recovers t...

This document introduces LLMFP, a framework for using Large Language Models (LLMs) to solve complex planning tasks by converting them into optimization problems solvable by formal planners like SMT solvers. It details the five-step process of LLMFP, which involves LLMs defining the problem, formulating variables, generating code, formatting results, and performing self-assessment and modification....

This paper introduces Automated Design of Agentic Systems (ADAS) , a new research area aiming to automatically invent building blocks and design powerful agentic systems , arguing that this could be a faster and more effective approach than manual design, drawing parallels to the history of machine learning where learned solutions surpassed hand-designed ones. The paper proposes and evaluates Meta...

This paper investigates the mathematical basis of large language model (LLM) prompting by framing LLMs as discrete stochastic dynamical systems and employing control theory. The authors formalize LLM systems and introduce concepts of controllability and reachability in this context. They present a Self-Attention Control Theorem that provides a theoretical limit on controlling self-attention output...

This academic paper from the University of Chicago addresses the problem of aligning large language models (LLMs) with human preferences. The authors analyze  best-of-n sampling , a technique where an LLM generates multiple responses and selects the best one, finding it to be  nearly optimal for maximizing win rate  while minimizing changes to other aspects of the output. To avoid the computationa...

This paper proposes a novel Bayesian inference perspective for understanding and improving fine-tuning methods for large language models (LMs). The authors argue that traditional Reinforcement Learning (RL) approaches, when applied naively, lead to distribution collapse , where the LM generates only a limited set of high-reward outputs. They demonstrate that the commonly used KL-regularized RL obj...

This academic paper investigates  language model alignment , a process of adjusting a base language model to better align with desired outcomes, often guided by a  reward model . It specifically examines two common alignment methods:  KL-constrained reinforcement learning (RL) , which maximizes reward while limiting divergence from the original model, and  best-of-N selection , where the highest-r...

We investigate how various reward signals, even  spurious  and  random  ones, impact the performance of different language models fine-tuned for mathematical reasoning using Reinforcement Learning from Verbose Reasoning (RLVR). The research demonstrates that while  Qwen  models show significant improvement even with  weak  or  incorrect  rewards, this benefit is not universal, with  Llama  and  OL...

This paper critically examines the  best-of-n policy , a common method for aligning generative language models by selecting the highest-reward sample from $n$ options drawn from a  reference policy . It  disproves a widely-used analytical formula  for the KL divergence between the best-of-n policy and the reference, proving that the formula is only an  upper bound . The authors analyze the conditi...

This research paper explores using score matching —a technique for estimating the gradient of a data distribution's logarithm—to perform causal discovery in nonlinear additive noise models . The authors demonstrate that the causal graph's structure can be inferred from the score function , particularly its Jacobian . They propose a new algorithm, SCORE , which estimates the causal order by...

This research paper focuses on  improving score-based generative models (SBGMs)  to produce high-quality, high-resolution images. The authors identify limitations of existing SBGMs, specifically their inability to scale to higher resolutions and occasional training instability. They propose  new theoretical analyses  and techniques for  selecting noise scales , an  efficient method for incorporati...

This paper introduces Disagreement-Aware Confidence Alignment (DACA) , an unsupervised method for calibrating the confidence of post-trained large language models (PoLMs) . While pre-trained language models (PLMs) are typically well-calibrated, post-training can lead to over-confidence , especially with limited labeled data. DACA addresses this by leveraging the well-calibrated confidence of PLMs...

This paper introduces AlphaEvolve, a system designed to  automate the discovery of advanced algorithms  by leveraging large language models (LLMs) within an evolutionary framework. The system works by taking a user-defined problem and evaluation criteria, then iteratively generating and improving code solutions through an evolutionary process powered by LLM ensembles. AlphaEvolve has successfully...

This academic paper presents vec2vec , a novel method for translating text embeddings between different models without requiring paired data or prior knowledge of the encoders. The authors demonstrate that this unsupervised technique successfully aligns embeddings from various models into a universal latent space , preserving the geometric structure and semantics of the original data. They show th...

This academic paper investigates how humans represent and infer goals , proposing that goals can be formalized as reward-producing programs . The researchers developed a physics-based game environment where participants created and demonstrated novel goals. By collecting natural language descriptions and formal scoring criteria , they analyzed the relationship between goal complexity, reward struc...

This paper introduce Trial-Error-Explain In-Context Learning (TICL) , a method for personalizing large language models (LLMs) to match individual user writing styles without requiring model fine-tuning. TICL expands the in-context learning prompt by adding model-generated negative examples and explanations that highlight discrepancies from the target style. Evaluations demonstrate that TICL signif...

This research investigates how little training data is needed for Reinforcement Learning with Verifiable Reward ( RLVR ) to significantly boost the mathematical reasoning abilities of  large language models (LLMs) . Surprisingly, the authors demonstrate that training on even just  one carefully chosen example  can achieve performance comparable to using datasets containing thousands, resulting in...

This paper introduces  Test-Time Reinforcement Learning (TTRL) , a novel method enabling Large Language Models (LLMs) to improve performance on unlabeled test data using Reinforcement Learning (RL). TTRL overcomes the lack of ground-truth labels by employing  majority voting on multiple model outputs to estimate rewards , essentially allowing models to self-supervise their training. The research d...

This paper presents research exploring whether a model-free reinforcement learning agent, specifically a DRC agent playing the game Sokoban , learns to plan. Through a concept-based interpretability methodology involving probing for planning-relevant concepts like future agent and box movements, investigating how plans are formed internally, and verifying the causal link between internal represent...

This paper proposes a new reward system for large language models (LLMs) called agentic reward modeling, which aims to create more reliable rewards by integrating human preferences with verifiable correctness signals . An empirical implementation, named REWARDAGENT , is presented, which combines human preference rewards with signals related to factuality and instruction following . Extensive exper...